Objective/Rationale:
Parkinson’s disease imaging using dopamine transporter (DAT) single-photon emission computed tomography (SPECT) technology conveys important information on brain dopamine activity at the pixel level. However, a commonly used method called regional intensity averaging oversimplifies the available information. In this project, researchers will explore a new way to measure dopamine transporter activity in Parkinson’s disease (PD) that applies shape and texture analyses to DAT SPECT images.
Project Description:
Scientists will analyze DAT scans from the Parkinson’s Progression Markers Initiative (PPMI) dataset. They will utilize a range of shape and texture analysis techniques, and investigate how the resolutions obtained in SPECT imaging impact the applicability of such techniques. Furthermore, data obtained from the various analysis methods will be combined using advanced mathematical algorithms to identity data combinations that are most important to characterize disease progression at various stages.
Relevance to Diagnosis/Treatment of Parkinson’s Disease:
The successful completion of this study could have important clinical implications. Enhanced sensitivity to subtle neuroanatomic changes is expected to provide novel insights into the relationship between dopaminergic alterations and PD manifestations. In addition to aiding identification of early disease, these techniques, if successful, can be applied to assess the impact of disease-modifying therapies.
Anticipated Outcome:
These researchers expect to demonstrate DAT change between early PD patients and healthy controls in the putamen, the most affected region. They also expect to demonstrate an ability to detect subtle abnormalities in dopaminergic function that might be associated with different manifestations of disease. Additionally, this work could enable future efforts towards understanding of subjects having scans without evidence of dopaminergic deficit (SWEDDs).
Final Outcome
We explored a range of feature extraction metrics and applied them to DaT SPECT images as available in the PPMI dataset. These metrics captured various different aspect of neurochemical tracer uptake, beyond conventional averaging, which oversimplifies the available information within the images. We found that some of our metrics significantly increased correlation of the imaging data with clinical scores and measures for PD subjects. In other words, our proposed paradigm is able to arrive at measures that better inform the relationship between dopaminergic alterations and PD manifestations. This approach also has potential to better assess the impact of novel disease-modifying therapies.
Presentations & Publications
S. Jain, Y. Salimpour, L. Younes, G. Smith, Z. Mari, V. Sossi, and A. Rahmim Application of pattern recognition framework for quantification of Parkinson's disease in DAT SPECT imaging Conference record: IEEE Nucl. Sci. Symp. Conf. Record, November 2014. Oral presentation.
S. A. L. Blinder, I. Klyuzhin, M. E. Gonzalez, A. Rahmim, and V. Sossi Texture and shape analysis on high and low spatial resolution emission images Conference record: IEEE Nucl. Sci. Symp. Conf. Record, November 2014. Oral presentation.
Y. Salimpour, S. Jain, G. Smith, Z. Mari, V. Sossi, and A. Rahmim Investigating the regional correlation of subcortical structures as imaged by DAT SPECT with clinical phenotypes in Parkinson's disease Submitted to the Society of Nuclear Medicine & Molecular Imaging (SNMMI) 2015 Annual Meeting.
K. Lahouel, E. Variani, S. Jain, A. Rahmim, and B. Jedynak C^{-1}mu: a classifier family motivated by the central limit theorem Submitted to the 2015 International Conference on Machine Learning.
March 2015